NMR Signal Processing with Wavelet Denoising

Spectroscopic imaging (i.e., the combination of nuclear magnetic resonance (NMR) spectroscopy with magnetic resonance imaging) has recently been accepted as a radiological procedure, recognized by health insurance companies. These studies typically use H NMR and quantify various metabolic biomarkers of disease. Recently, pattern recognition algorithms have been applied. The primary problem with applying pattern recognition algorithms to in vivo NMR data is poor spectral resolution. The advantage of genomic studies is that genetically modified animals allow the use of significantly higher magnetic field strengths, which increase spectral resolution. In addition, use of C NMR rather than H NMR increases spectral dispersion by an order of magnitude. However, this increase in resolution in C NMR as compared to H NMR is matched by a decrease in sensitivity. Therefore, new signal processing methods are needed to eliminate noise in in vivo C NMR spectra. In this study, in vivo and in vitro C NMR spectra were obtained from liver of intact rat and the corresponding perchloric acid hepatic extract, respectively, and wavelet denoising was applied to enhance spectral resolution and sensitivity.